Research

Implementation research on community health workers’ provision of maternal and child health services in rural Liberia Peter W Luckow,a Avi Kenny,b Emily White,b Madeleine Ballard,c Lorenzo Dorr,b Kirby Erlandson,d Benjamin Grant,b Alice Johnson,b Breanna Lorenzen,e Subarna Mukherjee,b E John Ly,b Abigail McDaniel,b Netus Nowine,f Vidiya Sathananthan,b Gerald A Sechler,g John D Kraemer,h Mark J Siedneri & Rajesh Panjabig

Objective To assess changes in the use of essential maternal and child health services in Konobo, Liberia, after implementation of an enhanced community health worker (CHW) programme. Methods The Liberian Ministry of Health partnered with Last Mile Health, a nongovernmental organization, to implement a pilot CHW programme with enhanced recruitment, training, supervision and compensation. To assess changes in maternal and child health-care use, we conducted repeated cross-sectional cluster surveys before (2012) and after (2015) programme implementation. Findings Between 2012 and 2015, 54 CHWs, seven peer supervisors and three clinical supervisors were trained to serve a population of 12 127 people in 44 communities. The regression-adjusted percentage of children receiving care from formal care providers increased by 60.1 (95% confidence interval, CI: 51.6 to 68.7) percentage points for diarrhoea, by 30.6 (95% CI: 20.5 to 40.7) for fever and by 51.2 (95% CI: 37.9 to 64.5) for acute respiratory infection. Facility-based delivery increased by 28.2 points (95% CI: 20.3 to 36.1). Facility-based delivery and formal sector care for acute respiratory infection and diarrhoea increased more in agricultural than gold-mining communities. Receipt of one-or-more antenatal care sessions at a health facility and postnatal care within 24 hours of delivery did not change significantly. Conclusion We identified significant increases in uptake of child and maternal health-care services from formal providers during the pilot CHW programme in remote rural Liberia. Clinic-based services, such as postnatal care, and services in specific settings, such as mining areas, require additional interventions to achieve optimal outcomes.

Introduction respiratory infection and malaria; and (ii) maternal and new- born care. Here, we describe the programmatic components, Over 95% of global maternal and child deaths occur in 75 low- implementation and an assessment of changes in maternal and and middle-income countries and remote populations within child health-care use three years after implementation. these countries often bear the greatest burden.1,2 Many countries are exploring strategies to scale up community health worker (CHW)-based programmes, which have been demonstrated to Methods improve health in the domains of maternal and child health, Setting and participants access to family planning and prevention of human immuno- deficiency virus (HIV) infection, malaria and tuberculosis.3,4 The programme took place in Konobo district in south-eastern In Liberia, an estimated 60% (1.2 million people) of the ru- Liberia. Konobo is one of Liberia’s most remote regions, com- ral population lives more than 5 km from the nearest health fa- prised of 2983 km2 of rainforest, with a population density of cility and the country has among the highest maternal and child 4.1 people/km2. In 2012, approximately 12 000 residents in mortality rates globally, 725 deaths per 100 000 live births and the district lived more than 5 km from the nearest clinic. One 70 deaths per 1000 live births, respectively.5–7 In 2012, the health quarter were women of reproductive age (15–49 years) and 16% ministry partnered with Last Mile Health, a nongovernmental were children under five years. These two demographic groups organization, to pilot a programme for enhanced CHW-based represented the target population of the programme. All 44 re- health care for remote populations (those living farther than mote communities in Konobo, located more than 5 km from the 5 km or a one hour walk from the nearest health facility). The district’s only health clinic were involved with the programme. programme aimed to increase coverage of essential maternal The average road distance from these communities to the clinic and child health services through enhanced recruitment, train- was approximately 25 km and the mean population of the ing, supervision and compensation of CHWs. Responsibilities communities was 276 people. According to the 2012 Liberian of CHWs included provision of: (i) integrated community case Demographic and Health Survey, Konobo had worse maternal management of childhood illnesses, including diarrhoea, acute and child health outcomes than other rural Liberian districts.8

a Geisel School of Medicine at Dartmouth College, Hanover, United States of America (USA). b Last Mile Health, Monrovia, Liberia. c Department of Social Policy and Intervention, University of Oxford, Oxford, England. d Harvard Medical School, Boston, USA. e University of Minnesota Medical School, Minneapolis, USA. f Grand Gedeh County Health Team, Ministry of Health, Monrovia, Liberia. g Division of Global Health Equity, Brigham and Women’s Hospital, Harvard Medical School, 75 Francis Street, Boston, MA, 02115, USA. h Department of Health Systems Administration, Georgetown School of Nursing and Health Studies, Washington, USA. i Department of Medicine, Harvard Medical School, Boston, USA. Correspondence to Rajesh Panjabi (email: [email protected]). (Submitted: 31 May 2016 – Revised version received: 7 October 2016 – Accepted: 9 October 2016 )

Bull World Health Organ 2017;95:113–120 | doi: http://dx.doi.org/10.2471/BLT.16.175513 113 Research Maternal and child care in Liberia Peter W Luckow et al.

Implementation to health clinics. A typical training took therapy. Malaria was diagnosed with two weeks to complete. Physician assis- rapid diagnostic tests in children, with a We implemented the programme be- tants and registered nurses led the training measured fever. Additionally, CHWs con- tween 2012 and 2015 in a stepwise fash- sessions that focused on clinical skills, ducted home-based antenatal care educa- ion over three geographic areas within such as history-taking, physical examina- tion, helped design birth plans, scheduled Konobo district. Integrated community tion and specific clinical procedures, such facility-based deliveries, screened preg- case management of childhood illness as rapid diagnostic testing for malaria. nant women and neonates for danger was launched in February 2013, August After the start of the Ebola virus disease signs, referred cases with danger signs 2013 and March 2014. Maternal and outbreak, we added a training module on to the clinic and promoted exclusive newborn care services were launched surveillance for Ebola symptoms. breastfeeding. CHWs provided services in November 2012, December 2013 and CHWs received weekly supervision at no cost to community members. To April 2015. visits from peer supervisors who were further promote health-care utilization, CHW recruitment and staffing trained in project and supply manage- CHWs organized community health ment, supportive supervision and refer- committees that partnered with trained We recruited CHWs through commu- rals. Supervision visits were designed to traditional midwives to refer expectant nity nomination, as recommended by last one hour each and consisted of form mothers to stay at maternal waiting homes the 2008 National policy and strategy on reviews, patient audits and restocking of (a residence near the clinic for at-term community health services.9 Our recruit- essential commodities. High-performing mothers), until childbirth. Beginning ment process added several components, CHWs were promoted to peer supervi- in April 2013, the programme paid the including (i) completion of a literacy sors, and were equipped with a motorbike midwives US$ 3–5 for clinic referrals, and test; (ii) an in-person interview to assess for travel during supervision visits. Ini- mothers who delivered in the clinic were motivation and communication skills; tially, the visits were unstructured and left provided transport reimbursements and and (iii) training and subsequent skills’ to the discretion of individual supervisors, food stipends. assessment for candidates that passed the however, since May 2013, supervision We modified several elements of the interview. After this additional three-step visits included the use of quality assur- programme during the Ebola virus disease screening and assessment, we identified ance checklists and randomly-sampled outbreak. After the start of the outbreak and hired the highest scoring CHWs. We patient audits led by the peer supervisors. in 2015, CHWs performed active surveil- also conducted a follow-up competency Separately, clinical supervisors conducted lance through monthly household visits. assessment during the first 90-days of monthly field supervision visits to assess While there were no confirmed cases their employment. We recruited CHWs adherence to clinical protocols and pro- of Ebola in the study area, use of rapid from the communities in which they vide formative feedback based on form malaria diagnostic tests was suspended resided, and they served communities reviews and direct observation of patient to ensure the safety of CHWs, and the within a 30-minute walk from their home interactions. programme adopted treatment protocols community. Although the 2008 National policy based on self-reported signs and symp- We also interviewed and hired two and strategy on community health ser- toms. Similar changes were implemented types of supervisors: clinical supervi- vices specified in-kind compensation for for treatment of diarrhoea and acute sors (i.e. clinic-based community health CHWs,9 an agreement with the health respiratory infection. nurses and physician assistants) and peer ministry allowed an additional monthly Data collection and analysis supervisors (i.e. non-clinical supervisors cash payment to CHWs and supervisors. who conduct process supervision and CHWs were paid 60 United States dol- We used data from two population-rep- community engagement). Best perform- lars (US$) per month for an estimated resentative household surveys conducted ing CHWs were promoted to serve as 20 hours of work per week, while peer by Last Mile Health in August 2012 and peer supervisors. supervisors and clinical supervisors were August 2015. Fundamental aspects of the paid US$ 150 and US$ 550 per month survey design and execution were de- Training, supervision and 2 compensation respectively. scribed previously. The questionnaire was Programme services adapted from the 2007 and 2012 Liberian CHWs completed an initial two-week Demographic and Health Surveys and training in the district capital. Training Initially, services were provided through included sections on household charac- modules focused on community leader- passive surveillance, whereby community teristics, maternal and neonatal health, re- ship (to promote community engage- members visited CHW households during productive health, child health and access ment), household mapping (to define his/ periods of illness or pregnancy. CHWs to health care. We used a two-stage cluster her catchment population) and registra- were trained to conduct integrated com- design for the sampling, which provided a tion (to assess demographics). Subsequent munity case management for diarrhoea, representative sample for assessing chang- modules were administered in the district acute respiratory infection and malaria, es in maternal and child health-care use. capital roughly once every three months along with referral of cases that presented We constructed a sampling frame using and focused on preventive and curative with danger signs. They were equipped raw data from the 2008 Liberian Census. components of maternal, neonatal and with diagnostic tools, including rapid The frame was adjusted using information child health services, including birth plan- malaria diagnostic tests, mid-upper arm from household enumeration performed ning, perinatal care, integrated community circumference bands and thermometers, by Last Mile Health before each survey. case management of malaria, acute respira- and therapeutics, including zinc, oral Communities were the primary sampling tory infection and diarrhoea, and criteria rehydration salts, amoxicillin, acetamino- units and were selected using probability- for referral of patients with warning signs phen and artemisinin-based combination proportional-to-size sampling. Individual

114 Bull World Health Organ 2017;95:113–120| doi: http://dx.doi.org/10.2471/BLT.16.175513 Research Peter W Luckow et al. Maternal and child care in Liberia households served as secondary sampling implementation was restricted to births CHWs and 10 supervisors, of whom units. Random selection of households captured after April 2013, when all catch- 39 CHWs and 5 supervisors remained within communities was done through ment communities had initiated at least active at the end of the pilot period. a random walk procedure. For the 2012 one maternal health programme element. By the completion of the programme, baseline survey, we had a total sample of Data analysis CHW-to-population ratio within the 600 households, selected from 30 clusters. study area was 1:311, which was more Last Mile Health updated the sampling We conducted descriptive analyses to than threefold higher than the ratio of frame for the 2015 follow-up survey after summarize respondent characteristics 1:1 000 proposed in the Liberian health a re-count of all the households in the at baseline and after programme imple- ministry’s policy.9 district. The 2015 sample included 1035 mentation. We fit logistic regression Respondent characteristics households, selected from 45 clusters. models to compare differences in each of We interviewed women ages 18–49 years the outcome indicators before and after Table 1 summarizes the survey respon- in both surveys. implementation. The regression models dents’ characteristics. We used data from We made certain changes to the sur- for maternal health were adjusted for 364 women of the 2012 survey and 205 vey between 2012 and 2015. We added community type (agricultural versus gold- women, who met the inclusion criteria questions to the follow-up survey on mining), maternal age, distance to health of the 2015 survey. The mean maternal asset ownership, family planning, pro- facility (measured by global positioning ages were 30 and 29 years (P = 0.002), vider use, vaccination and knowledge of system) and presence or absence of mo- respectively. Completing secondary Ebola. Before and after implementation, tor vehicle access to the nearest health education was more common in the comparisons were restricted to consistent facility. The models for child health were follow-up survey (P = 0.039). There were items between surveys and to communi- adjusted for these same variables as well no statistically significant differences ties common to both sampling frames. as the child’s age. After regression, we in other demographic characteristics. Individual weights for survey variables used predictive margins, holding covari- The number of children who met the were adjusted post-hoc based on 2015 ates at their observed values to estimate inclusion criteria was 470 in 2012 and data. In 2012, enumerators interviewed adjusted percentages of each outcome 452 in 2015. the woman in the household who most indicator before and after programme Child health recently completed a pregnancy, while in implementation, and tested before-to- 2015, all women within a household were after changes using contrasts of predicted Between 2012 and 2015, the proportion sampled and interviewed. The compara- percentages. Since the CHW programme of children receiving health care for tive analysis was therefore restricted to the had not started at the time of the base- childhood illnesses from formal provid- household woman who in the 2015 survey line survey in 2012, we estimated the ers significantly increased (Table 2). The responded as giving birth most recently. percentage of child health encounters for adjusted percentage increased by 60.1 Surveys were done in Liberian English and integrated community case management points (95% confidence interval, CI: 51.6 Konobo Krahn by bilingual enumerators. of childhood illnesses that were provided to 68.7) for diarrhoea, by 30.6 points Outcome measures by a CHW only for 2015. To assess mod- (95% CI: 20.5 to 40.7) for fever and by erating effects of community type, we 51.2 points (95% CI: 37.9 to 64.5) for We defined the child health-care use ran the same maternal and child health acute respiratory infection. Among those outcomes as management of childhood models with an interaction term of com- children who received formal provider illnesses by a formal care provider within munity type and programme period. All care in 2015, 83.5% (CI: 74.4 to 89.7) a two-week recall period. The childhood analyses incorporated complex sampling was from a CHW for diarrhoea, 78.6% illnesses included were: (i) diarrhoea; design using inverse probability weights (95% CI: 64.9 to 87.9) for acute respira- (ii) acute respiratory infection (defined and finite population corrections at both tory infection and 80.9% (95% CI: 73.3 as the combination of fever with a rapid stages. Standard errors were adjusted to 86.7) for fever. Formal sector care for respiratory rate); and (iii) fever. We de- for clustering using Taylor linearization. diarrhoea and fever increased more for fined formal care providers as community Statistical analyses were conducted us- children in agricultural than gold-mining health workers, ministry of health com- ing Stata version 14.2 (Statacorp, College communities, but the difference was not munity health volunteers (who were active Station, Texas, United States of America). statistically significant for acute respira- in some parts of Liberia but not in our Ethical considerations tory infection (Table 3). study area) and clinic staff. For maternal Maternal health and neonatal health-care use, we defined We obtained ethical approval for the sur- outcomes as: (i) completing at least one veys from the institutional review boards The adjusted facility-based delivery per- antenatal care visit at a health facility; of Partners Healthcare, Georgetown centage increased by 28.2 points (95% (ii) having a facility-based delivery; and University and the Liberian Institute for CI: 20.3 to 36.1; Table 2). The increase (iii) receiving postnatal care from a clinic Biomedical Research. Respondents gave was 39.0 percentage points (95% CI: staff member or a CHW within 24 hours of verbal informed consent. 29.8 to 48.2) in agricultural communities delivery. To assess change in child health- compared to 19.6 points (95% CI: 7.6 to care use, data from the 2012 survey and 31.5) in mining communities. There were the 2015 survey were used as before and Results no significant changes in the receipt of after programme implementation, respec- Implementation of programme at least one formal provider-associated tively. To assess changes with maternal antenatal care visit or receipt of postnatal care use, the 2012 survey was used as a Between October 2012 and August 2015, care within 24 hours. baseline, but assessment of programme we recruited and trained a total of 54

Bull World Health Organ 2017;95:113–120| doi: http://dx.doi.org/10.2471/BLT.16.175513 115 Research Maternal and child care in Liberia Peter W Luckow et al.

Table 1. Demographic characteristics of respondents and description of the survey site, by survey year, Konobo district, Liberia, 2012 and 2015

Characteristic Unweighteda Weightedb 2012 No. (%) 2015 No. (%) 2012 2015 (n = 364) (n = 205) % (95% CI) % (95% CI) (n = 364) (n = 205) Maternal age 30.5 (N/A)c 28.6 (N/A)c 30.4c (29.6 to 31.1) 28.5c (27.7 to 29.4) Maternal education None 99 (27.6) 61 (27.8) 24.8 (20.4 to 29.0) 30.8 (25.5 to 36.7) Primary 211 (58.8) 117 (57.1) 54.8 (48.2 to 61.2) 57.0 (51.0 to 62.8) Secondary or higher 49 (13.7) 27 (13.2) 20.4 (14.4 to 28.8) 12.2 (9.0 to 16.4) Residents living in a gold mining community 119 (32.7) 121 (59.0) 58.1 (45.1 to 70.1) 58.2 (49.2 to 66.8) Distance to clinic 26.4 (N/A)d 26.1 (N/A)d 28.4d (26.6 to 30.2) 26.9d (24.7 to 29.1) Community accessible by vehicle 256 (70.3) 92 (44.9) 47.6 (34.3 to 61.3) 46.5 (37.8 to 55.5) CI: confidence interval; N/A: not applicable. c Reported in mean years. a Unweighted counts, means and percentages describe the sample only. d Reported in mean km. b Weighted values represent the population and are calculated using sampling weights.

Table 2. Change in maternal and child health-care use, before and after programme implementation, Konobo district, Liberia, 2012 and 2015

Outcome Before 2012 After 2015 Percentage Adjusted % (95% CI)a Adjusted % (95% CI)a point difference (95% CI) Child health-care use outcomes Diarrhoea treatment from formal provider 6.1 (3.6 to 10.3) 66.3 (56.9 to 74.5) 60.1 (51.6 to 68.7) Acute respiratory infection treatment from formal 6.6 (3.7 to 11.4) 57.8 (44.4 to 70.1) 51.2 (37.9 to 64.5) provider Fever treatment from formal provider 26.2 (20.5 to 33.0) 56.8 (49.3 to 64.1) 30.6 (20.5 to 40.7) Maternal health-care use outcomes Facility-based delivery 55.8 (49.2 to 62.3) 84.0 (79.1 to 88.0) 28.2 (20.3 to 36.1) One-or-more antenatal care sessions at a health facility 81.4 (76.6 to 85.4) 82.8 (77.5 to 87.0) 1.4 (−4.7 to 7.5) Postnatal care (maternal or neonatal) within 24 hours 17.1 (13.0 to 22.1) 19.4 (15.0 to 24.6) 2.3 (−4.2 to 8.8) from a clinic staff member or CHW CI: confidence interval; CHW: community health worker. a Adjusted values were produced using predictive margins after fitting multivariable logistic regression models. Note: Formal care providers included community health workers, ministry of health community health volunteers (who were active in some parts of Liberia but not the pilot programme catchment area) and clinic staff.

Discussion generally positive. CHW programmes lack of improvement. The low rates of improved care seeking for childhood postnatal care could be explained by a Here we evaluate, over a three-year pe- illnesses, though effectiveness and effect combination of the community focus of riod, the implementation of a programme sizes vary between interventions.3 Simi- the programme and Ebola-related effects recruiting and training CHWs to deliver larly, most high-quality impact evalu- on care seeking. However, postnatal care maternal and child care. Despite the Ebola ations report that CHW programmes receipt was lower than facility-based de- virus disease outbreak, which caused sub- improve child mortality, but results are liveries, which suggests missed opportu- stantial declines in health-care utilization mixed and individual studies are often nities at clinics. These results underscore in other regions of the country,10–13 we limited because of their relatively small the importance of integrating community show increases in health-care use from sample size.14,16 Systematic reviews and facility-based services throughout formal providers for fever, acute respira- have found substantial heterogeneity in the continuum of care.18,19 Additionally, tory infection and diarrhoea among chil- CHW programme components and ef- community-based postnatal services may dren and facility-based delivery among fects,3,14,16,17 suggesting a need for more be needed to increase postnatal care rates pregnant women. Our three-year follow- research on specific programme elements in remote locations with weak facility- up period is longer than many prior evalu- and across contexts. based services.20,21 ations.14 While many studies do not report We did not detect significant im- Programme efficacy was generally distance to clinic, we did not identify any provements in rates of antenatal or lower in gold-mining communities than studies from areas as remote as Konobo.15 postnatal care. Antenatal care rates were agricultural communities. In Konobo, Previous CHW programme evalu- already high at baseline (over 80% of mining communities are usually larger ations report mixed findings, but are pregnancies), which could explain the and more transient. Limited evidence

116 Bull World Health Organ 2017;95:113–120| doi: http://dx.doi.org/10.2471/BLT.16.175513 Research Peter W Luckow et al. Maternal and child care in Liberia

suggests that CHWs function better with high social capital,22–24 a contextual mod- erator that is likely reduced in mining communities. Additionally, mining com- munities tend to have greater availability of private-sector pharmaceutical ser- vices. Studies have shown that alternative 0.7 (−7.9 to 9.2) 0.7 (−7.9 to 2.7 (−6.6 to 11.9) 2.7 (−6.6 to 9.7 (−4.6 to 23.9) 9.7 (−4.6 to Percentage point Percentage 19.6 (7.6 to 31.5) 19.6 (7.6 to 46.8 (23.9 to 69.7) 46.8 (23.9 to 55.4 (43.6 to 67.1) 55.4 (43.6 to difference (95% CI) difference suppliers replace formal-sector services, particularly when transportation is costly or facility-based services are perceived to a be of low quality.25,26 Future investigation will need to assess the sustainability and scalability of these programmes. In addition to sus- After 2015 After tainable funding, pilot programmes often Mining communities 15.5 (9.9 to 23.6) 15.5 (9.9 to 82.1 (73.3 to 88.4) 82.1 (73.3 to 80.6 (72.6 to 86.7) 80.6 (72.6 to 42.8 (30.6 to 56.0) 42.8 (30.6 to 48.7 (28.3 to 69.6) 48.7 (28.3 to 59.8 (47.0 to 71.4) 59.8 (47.0 to

Adjusted % (95% CI) Adjusted require alterations to remain appropriate for a wide variety of contexts as they get a expanded.27–29 Practices from this pro- gramme are being scaled up to over 240 remote communities in adjacent River- cess County. Furthermore, several of the

Before 2012 Before programme’s features, such as contracts 2.0 (0.3 to 10.2) 2.0 (0.3 to 4.4 (1.7 to 11.3) 4.4 (1.7 to 14.9 (9.3 to 22.9) 14.9 (9.3 to 79.4 (70.6 to 86.0) 79.4 (70.6 to 61.0 (50.3 to 70.8) 61.0 (50.3 to 33.1 (23.2 to 44.8) 33.1 (23.2 to and cash payments, ensuring a CHW- Adjusted % (95% CI) Adjusted to-population ratio of 1:350, targeting of services to remote communities and field-based supervision, have helped to inform the design of Liberia’s National Community Health Assistant Program.5 This programme was launched in 2016 to accelerate progress towards universal 4.8 (−4.2 to 13.9) 4.8 (−4.2 to 0.1 (−7.7 to 7.8) 0.1 (−7.7 to Percentage point Percentage 39.0 (29.8 to 48.2) 39.0 (29.8 to 61.2 (50.8 to 71.5) 61.2 (50.8 to 61.7 (41.8 to 81.7) 61.7 (41.8 to 79.7 (66.4 to 93.0) 79.7 (66.4 to health coverage for the most vulnerable difference (95% CI) difference populations, especially those in remote communities.5 The newly launched pro- a gramme seeks to transform an existing cadre of unpaid and poorly coordinated CHWs into a more effective workforce by enhancing recruitment, supervision After 2015 After and compensation. The health ministry 25.8 (16.2 to 38.7) 25.8 (16.2 to 83.5 (74.3 to 89.9) 83.5 (74.3 to 87.8 (79.7.to 92.9) 87.8 (79.7.to 80.6 (67.4 to 89.3) 80.6 (67.4 to 78.0 (48.7 to 93.0) 78.0 (48.7 to 88.2 (66.9 to 96.5) 88.2 (66.9 to

Adjusted % (95% CI) Adjusted has organized a coalition of funding and

Agricultural communities implementation partners to support this

a new programme. Formal evaluations of both effectiveness and cost–effectiveness are planned as part of the scale-up. Our study has several limitations. First, results are uncontrolled, limit- Before 2012 Before

8.5 (4.0 to 17.1) 8.5 (4.0 to ing causal inferences. However, we are 21.0 (13.9 to 30.6) 21.0 (13.9 to 83.4 (77.5 to 88.1) 83.4 (77.5 to 48.8 (39.7 to 57.9) 48.8 (39.7 to 19.4 (12.9 to 28.1) 19.4 (12.9 to 16.3 (5.6 to 39.2) 16.3 (5.6 to

Adjusted % (95% CI) Adjusted unaware of any other programmes that occurred during implementation and our data show that by 2015 over 75% of child health services were reported to be delivered by CHWs, lending sup- port to a causal inference.30 Second, we cannot differentiate the effects of the CHW programme from the effects of the trained traditional midwives’ incen- tives, transport reimbursements and food stipends, which were simultaneously implemented. Similarly, we cannot iden-

Change in maternal and child health-care use before and after programme implementation, by community type, Konobo district, Liberia, 2012 and 2015 community type, and child health-care by Change implementation, programme in maternal and after use before tify the independent effects of particular CHW programme sub-elements, such as

Adjusted values were produced using predictive margins after fitting multivariable logistic regression models. regression after fitting multivariable logistic margins using predictive produced were values Adjusted supervision versus compensation. Third,

a CI: confidence interval; CHW: community worker. interval; health CI: confidence CHW: Note: Formal care providers included community health workers, ministry of health community health volunteers (who were active in some parts of Liberia but not the pilot programme catchment area) and clinic staff. area) active in some parts ministry catchment included community health workers, (who were of Liberia providers but not the pilot programme of health community health volunteers care Formal Note: Table 3. Table Postnatal care (maternal or neonatal) within 24 or neonatal) (maternal care Postnatal a clinic staff member or CHW hours by One-or-more antenatal care sessions at a health sessions at One-or-more care antenatal facility Maternal health-careMaternal use outcomes delivery Facility-based Fever treatment from formal provider formal from treatment Fever Acute respiratory infection treatment from from infection respiratory treatment Acute provider formal Outcome Child health-care use outcomes provider formal from Diarrhoea treatment our programme was done in a single dis-

Bull World Health Organ 2017;95:113–120| doi: http://dx.doi.org/10.2471/BLT.16.175513 117 Research Maternal and child care in Liberia Peter W Luckow et al. trict with a small population, therefore, Tolbert Nyenswah, Tamba Boima, and The Lester Fund. RP received support the results are not generalizable for all the Grand Gedeh, Rivercess County from the Harvard Burke Global Health remote populations. Health Teams of Liberia’s Ministry of Fellowship. MJS receives support from This paper offers preliminary data Health, Fiona Walsh, Lisha McCormick, the National Institutes of Health (MH on how an enhanced CHW-based pro- Bakary Sidibe, Michael Zouzoua and K23099916) and the Harvard Center for gramme was used to promote the uptake their teams at Last Mile Health in Boston, AIDS Research (5P30AI060354). of essential maternal and child health New York, Monrovia, Grand Gedeh and services in remote populations. Future Rivercess. Competing interests: None declared. investigations will assess the sustainabil- ity and scalability of the programme. ■ Funding: Funding for the surveys used to collect data was provided, in part, by Acknowledgements the UBS Optimus Foundation, Direct We thank Bernice Dahn, Francis Kateh, Relief, The Greenbaum Foundation, Walter Gwenigale, Samson K Arzuaquoi, the Global Neighborhood Fund and

ملخص بحث حول التنفيذ العميل لربنامج تقديم اخلدمات الصحية لألمهات واألطفال من جانب األخصائيني الصحيني للمجتمع املحيل يف املناطق الريفية من ليبرييا تقييم الغرضالتغيريات التي طرأت عىل االستعانة باخلدمات 20.5 إىل 40.7( نقطة مئوية للحمى، وبمقدار 51.2 )بمستوى الرضورية لصحة األمهات واألطفال يف كونوبو بليبرييا بعد تنفيذ ثقة تبلغ نسبته 95%: 37.9 إىل 64.5( للعدوى التنفسية احلادة. برنامج ّ ن حمسلألخصائيني الصحيني للمجتمع املحيل. كام تزايد إجراء عمليات الوالدة يف املرافق الصحية بمقدار 28.2 دخلتالطريقة وزارة الصحة الليبريية يف إطار من الرشاكة مع نقطة )بمستوى ثقة تبلغ نسبته 95%: 20.3 إىل 36.1(. وقد منظمة “الست مايل هيلث” الصحية غري احلكومية لتنفيذ برنامج حتققتزيادة أكرب يف نسبة عمليات الوالدة يف املرافق الصحية جتريبي لألخصائيني الصحيني للمجتمع املحيل مع تطوير عمليات والرعاية املقدمة من جانب اجلهات الرسمية حلاالت العدوى التعيني والتدريب واإلرشاف وتقديم األجور. ولكي يتم تقييم احلادة يف اجلهاز التنفيس واإلسهال يف املجتمعات املحلية الزراعية التغيريات يف جمال االستعانة بالرعاية الصحية لألمهات واألطفال، باملقارنة مع املجتمعات املحلية القائمة عىل تعدين الذهب. ومل تطرأ فقد أجريت جمموعة من االستبيانات املجمعة القطاعية املتكررة زيادة ملموسة عىل تلقي جلسات الرعاية ملرحلة ما قبل الوالدة قبل تنفيذ الربنامج )يف عام (2012 وبعد تنفيذه )يف عام 2015(. بمعدل جلسة واحدة أو أكثر يف املرافق الصحية وخدمات الرعاية يف النتائجالفرتة ما بني ّعامي 2012 و2015 تم تدريب 54 بعد الوالدة خالل فرتة 24ساعة من الوالدة. من األخصائيني الصحيني للمجتمع املحيل، وسبعة مرشفني من لقداالستنتاج حددنا زيادات ملموسة يف حجم تلقي خدمات النظراء، وثالثة مرشفني رسيريني لتقديم اخلدمة لرشحية سكانية الرعاية الصحية لألمهات واألطفال من اجلهات الرسمية خالل يبلغ قوامها 12،127 نسمة يف 44 ًجمتمعا ًحمليا. وقد زادت فرتة الربنامج التجريبي لألخصائيني الصحيني للمجتمع املحيل النسبة املئوية ّاملعدلة ًا وفقلنموذج االنحدار اإلحصائي لألطفال يف املناطق الريفية النائية يف ليبرييا. وحتتاج اخلدمات املعتمدة عىل الذين يتلقون الرعاية من مقدمي الرعاية الرسميني بمقدار 60.1 العيادات – مثل الرعاية يف مرحلة ما بعد الوالدة - واخلدمات )بمستوى ثقة تبلغ نسبته 95%: 51.6 إىل 68.7( نقطة مئوية املقدمة يف بيئات حمددة، مثل مناطق التعدين إىل تدخالت إضافية بالنسبة لإلسهال، وبمقدار 30.6 )بمستوى ثقة تبلغ نسبته 95 %: لتحقيق أفضل النتائج.

摘要 利比里亚农村地区社区卫生工作者提供妇幼卫生服务的实施研究 目的 旨在评估实施加强的社区卫生工作者 (CHW) 计 个百分点，发热儿童百分比提高了 30.6 划以后，利比里亚科诺博地区基本妇幼卫生服务使用 （95% CI ： 20.5 至 40.7），急性呼吸道感染儿童的百分 情况的变化。 比提高了 51.2（ 95% CI ： 37.9 至 64.5）。 入院分娩提高 方法 利比里亚卫生部联手最后一里医疗 (Last Mile 了 28.2 个百分点（95% CI ： 20.3 至 36.1）。 与采金地 Health)，一家非政府组织，实施在招募、培训、监督 区相比，农业地区入院分娩和前往正规科室治疗急性 和报酬方面均有所加强的社区卫生工作者 (CHW) 试 呼吸道感染和腹泻的情况增加得更多。 在医疗机构接 点计划。 为了评估妇幼保健使用的变化，我们在该计 受一项或多项产前护理课程和产后 24 小时内护理的 划实施之前 (2012) 和之后 (2015) 分别开展了重复性横 情况没有显著变化。 断面群体调查。 结论 我们发现利比里亚偏远农村地区在社区卫生工作 结果 2012 至 2015 年间，54 名社区卫生工作者、七 者 (CHW) 计划试点期间从正规医疗机构接受妇幼保 名同伴监管员和三名临床监管员接受了培训，以 健服务的情况显著增加。 腹泻护理等临床服务和采矿 服 务 44 个社区内的 12 127 名居民。 从正规护理 地区等特定环境下的服务，需要额外的干预才能达到 提供者处接受护理的腹泻儿童的回归调整百分 最理想的效果。 比提高了 60.1【95% 置信区间 (CI)： 51.6 至 68.7】

118 Bull World Health Organ 2017;95:113–120| doi: http://dx.doi.org/10.2471/BLT.16.175513 Research Peter W Luckow et al. Maternal and child care in Liberia

Résumé Recherche opérationnelle sur la prestation de services de santé maternelle et infantile par des agents de santé communautaires dans les régions rurales du Libéria Objectif Évaluer les changements dans le recours aux services essentiels fièvre et de 51,2 (IC 95%: 37,9 à 64,5) pour les infections aiguës des de santé maternelle et infantile à Konobo, au Libéria, après la mise en voies respiratoires. Les accouchements en maternité ont augmenté de œuvre d’un programme de perfectionnement des agents de santé 28,2 points (IC 95%: 20,3 à 36,1). Les accouchements en maternité et la communautaires. prise en charge dans le secteur formel des infections aiguës des voies Méthodes Le ministère de la Santé libérien s’est associé avec Last Mile respiratoires et des cas de diarrhée ont davantage augmenté dans les Health, une organisation non gouvernementale, afin de mettre en œuvre communautés agricoles que dans celles vivant de l’extraction de l’or. un programme pilote destiné à améliorer le recrutement, la formation, Nous n’avons pas observé de changements significatifs dans le fait de l’encadrement et la rémunération des agents de santé communautaires. bénéficier d’une ou plusieurs séances de soins prénataux dans une Pour évaluer les changements au niveau des soins de santé maternelle structure de soins ou de soins post-partum dans les 24 heures suivant et infantile, nous avons réalisé plusieurs sondages transversaux par l’accouchement. grappes avant (2012) et après (2015) la mise en œuvre du programme. Conclusion Nous avons observé une augmentation significative Résultats Entre 2012 et 2015, 54 agents de santé communautaires, de la prestation de services de santé maternelle et infantile par des sept collègues superviseurs et trois superviseurs cliniques ont été prestataires officiels lors du programme pilote destiné aux agents formés pour prendre en charge une population de 12 127 personnes de santé communautaires dans les régions rurales isolées du Libéria. dans 44 communautés. Le pourcentage corrigé par régression d’enfants Des interventions supplémentaires sont nécessaires pour obtenir des recevant des soins de la part de prestataires de soins officiels a augmenté résultats optimaux vis-à-vis de la prestation de services cliniques, comme de 60,1 (intervalle de confiance (IC) de 95%: 51,6 à 68,7) points de les soins post-partum, et de services dans des zones spécifiques, comme pourcentage pour la diarrhée, de 30,6 (IC 95%: 20,5 à 40,7) pour la les régions minières.

Резюме Оказание услуг в сфере охраны здоровья матерей и детей местными медработниками в сельской местности Либерии: исследование в области внедрения Цель Оценить изменения в использовании основных услуг 20,5–40,7) при лихорадке и на 51,2 (95% ДИ: 37,9–64,5) при острых в сфере охраны здоровья матерей и детей в округе Конобо, респираторных инфекциях. Доля родов, принятых в учреждениях, Либерия, после внедрения усовершенствованной программы увеличилась на 28,2 пункта (95% ДИ: 20,3–36,1). Доля родов, для местных медицинских работников (ММР). принятых в учреждениях, и помощи, оказанной со стороны Методы Министерство здравоохранения Либерии вступило в официального сектора при острых респираторных заболеваниях партнерские отношения с неправительственной организацией и диарее, увеличилась больше в сельскохозяйственных общинах, Last Mile Health для внедрения пробной программы ММР, чем в золотодобывающих. Изменения в доле получивших предполагающей усовершенствования в наборе кадров, дородовую медицинскую помощь в рамках одного или обучении, надзоре и оплате труда. Чтобы оценить изменения в нескольких сеансов в медицинском учреждении и послеродовой использовании услуг в сфере охраны здоровья матерей и детей, уход в течение 24 часов после родов не были статистически авторы провели несколько межсекторальных обследований с значимы. применением гнездовой выборки до (2012 год) и после (2015 год) Вывод Авторы выявили существенное увеличение в внедрения программы. использовании услуг в сфере охраны здоровья матерей и Результаты В период между 2012 и 2015 годами было подготовлено детей, оказываемых официальными медицинскими работниками, 54 ММР, семь инспекторов из партнерских организаций и три в ходе реализации пробной программы ММР в удаленных инспектора из клиник для обслуживания 12 127 человек в сельских регионах Либерии. Клинические услуги, такие как 44 сообществах. Скорректированная, рассчитанная с помощью послеродовой уход, и обслуживание в специфических условиях, модели регрессии доля детей, которые получили медицинскую таких как районы разработки полезных ископаемых, требуют помощь от официальных работников здравоохранения, осуществления дополнительных вмешательств для достижения увеличилась на 60,1 (95% доверительный интервал, ДИ: 51,6– оптимального результата. 68,7) процентного пункта в случае с диареей, на 30,6 (95% ДИ:

Resumen Investigaciones sobre la implementación en el suministro de servicios de salud materna e infantil de trabajadores comunitarios de salud en la Liberia rural Objetivo Evaluar los cambios en el uso de servicios de salud materna atención sanitaria materna e infantil, se realizaron repetidas encuestas e infantil básicos en Konobo, Liberia, tras la implementación de un transversales de conglomerados antes (2012) y después (2015) de la programa mejorado de trabajadores comunitarios de salud (CHW, por implementación del programa. sus siglas en inglés). Resultados Entre 2012 y 2015, 54 CHW, 7 supervisores homólogos Métodos El Ministerio de Salud de Liberia se asoció con Last Mile y 3 supervisores clínicos recibieron formación para trabajar para una Health, una organización no gubernamental, para implementar un población de 12 127 personas en 44 comunidades. El porcentaje programa piloto de CHW con una mejora en la contratación, formación, ajustado a la regresión de niños que recibieron atención de profesionales supervisión y compensación. Para evaluar los cambios en el uso de la sanitarios formales aumentó un 60,1% (intervalo de confianza (IC) del

Bull World Health Organ 2017;95:113–120| doi: http://dx.doi.org/10.2471/BLT.16.175513 119 Research Maternal and child care in Liberia Peter W Luckow et al.

95%: 51,6 a 68,7) para la diarrea, un 30,6% (IC del 95%: 20,5 a 40,7) Conclusión Se identificaron aumentos significativos en la aceptación para la fiebre y un 51,2% (IC del 95%: 37,9 a 64,5) para la infección de los servicios de atención sanitaria materna e infantil de profesionales respiratoria aguda. El suministro en centros aumentó un 28,2% (IC del formales durante la versión piloto del programa de CHW en la Liberia 95%: 20,3 a 36,1). El suministro en centros y la atención del sector formal rural remota. Los servicios clínicos, como la atención posparto y para la infección respiratoria aguda y la diarrea aumentaron más en los servicios en lugares concretos, como zonas mineras, requieren comunidades agrícolas que en las de minas de oro. La recepción de una intervenciones adicionales para lograr unos resultados óptimos. o más sesiones de atención prenatal en un centro sanitario y de atención posparto a las 24 horas del parto no experimentó cambios importantes.

References 1. Bhutta ZA, Black RE. Global maternal, newborn, and child health–so near 16. Christopher JB, Le May A, Lewin S, Ross DA. Thirty years after Alma-Ata: a and yet so far. N Engl J Med. 2013 Dec 5;369(23):2226–35. doi: http://dx.doi. systematic review of the impact of community health workers delivering org/10.1056/NEJMra1111853 PMID: 24304052 curative interventions against malaria, pneumonia and diarrhoea on child 2. Kenny A, Basu G, Ballard M, Griffiths T, Kentoffio K, Niyonzima JB, et al. mortality and morbidity in sub-Saharan Africa. Hum Resour Health. 2011 Remoteness and maternal and child health service utilization in rural Oct 24;9(1):27. doi: http://dx.doi.org/10.1186/1478-4491-9-27 PMID: Liberia: A population-based survey. J Glob Health. 2015 Dec;5(2):020401. 22024435 doi: http://dx.doi.org/10.7189/jogh.05.020401 PMID: 26207180 17. Okwundu CI, Nagpal S, Musekiwa A, Sinclair D. Home- or community-based 3. Lewin S, Munabi-Babigumira S, Glenton C, Daniels K, Bosch-Capblanch X, programmes for treating malaria. Cochrane Database Syst Rev. 2013 May van Wyk BE, et al. Lay health workers in primary and community health care 31; (5):CD009527. PMID: 23728693 for maternal and child health and the management of infectious diseases. 18. Kerber KJ, de Graft-Johnson JE, Bhutta ZA, Okong P, Starrs A, Lawn JE. Cochrane Database Syst Rev. 2010 Mar 17; (3):CD004015. doi: http://dx.doi. Continuum of care for maternal, newborn, and child health: from slogan to org/10.1002/14651858.CD004015.pub3 PMID: 20238326 service delivery. Lancet. 2007 Oct 13;370(9595):1358–69. doi: http://dx.doi. 4. Perry HB, Zulliger R, Rogers MM. Community health workers in org/10.1016/S0140-6736(07)61578-5 PMID: 17933651 low-, middle-, and high-income countries: an overview of their 19. Koblinsky M, Matthews Z, Hussein J, Mavalankar D, Mridha MK, Anwar I, history, recent evolution, and current effectiveness. Annu Rev Public et al.; Lancet Maternal Survival Series steering group. Going to scale with Health. 2014;35(1):399–421. doi: http://dx.doi.org/10.1146/annurev- professional skilled care. Lancet. 2006 Oct 14;368(9544):1377–86. doi: publhealth-032013-182354 PMID: 24387091 http://dx.doi.org/10.1016/S0140-6736(06)69382-3 PMID: 17046470 5. Revised national community health services policy 2016–2021. Monrovia: 20. Aboubaker S, Qazi S, Wolfheim C, Oyegoke A, Bahl R. Community health Ministry of Health; 2015. workers: A crucial role in newborn health care and survival. J Glob Health. 6. Trends in maternal mortality: 1990 to 2015. Geneva: World Health 2014 Dec;4(2):020302. doi: http://dx.doi.org/10.7189/jogh.04.020302 PMID: Organization; 2015. 25520788 7. Levels and trends in child mortality 2015: estimates developed by the UN 21. McConnell M, Ettenger A, Rothschild CW, Muigai F, Cohen J. Can a Inter-agency group for child mortality estimation. New York: United Nations community health worker administered postnatal checklist increase health- Children’s Fund; 2015. seeking behaviors and knowledge?: evidence from a randomized trial 8. Kentoffio K, Kraemer JD, Griffiths T, Kenny A, Panjabi R, Sechler GA, et al. with a private maternity facility in Kiambu County, Kenya. BMC Pregnancy Charting health system reconstruction in post-war Liberia: a comparison Childbirth. 2016 Jun 04;16(1):136. doi: http://dx.doi.org/10.1186/s12884- of rural vs. remote healthcare utilization. BMC Health Serv Res. 2016 Sep 016-0914-z PMID: 27260500 07;16(1):478. doi: http://dx.doi.org/10.1186/s12913-016-1709-7 PMID: 22. Sato Y, Pongvongsa T, Nonaka D, Kounnavong S, Nansounthavong P, Moji K, 27604708 et al. Village health volunteers’ social capital related to their performance in 9. National policy and strategy on community health services: Monrovia: Lao People’s Democratic Republic: a cross-sectional study. BMC Health Serv Liberian Ministry of Health; 2008. Available from: http://liberiamohsw.org/ Res. 2014 Mar 12;14(1):123. doi: http://dx.doi.org/10.1186/1472-6963-14- Policies%20&%20Plans/Community%20Health%20Services%20Policy.pdf 123 PMID: 24620729 [cited 2016 Nov 11]. 23. Giblin PT. Effective utilization and evaluation of indigenous health care 10. Ly J, Sathananthan V, Griffiths T, Kanjee Z, Kenny A, Gordon N, et al. Facility- workers. Public Health Rep. 1989 Jul-Aug;104(4):361–8. PMID: 2502807 based delivery during the ebola virus disease epidemic in rural Liberia: 24. Leban K. How social capital in community systems strengthens in health analysis from a cross-sectional, population-based household survey. PLoS systems: people, structures, processes. USAID, CORE Group, MCHIP; 2011. Med. 2016 Aug 02;13(8):e1002096. doi: http://dx.doi.org/10.1371/journal. 25. Ndyomugyenyi R, Neema S, Magnussen P. The use of formal and informal pmed.1002096 PMID: 27482706 services for antenatal care and malaria treatment in rural Uganda. Health 11. Iyengar P, Kerber K, Howe CJ, Dahn B. Services for mothers and newborns Policy Plan. 1998 Mar;13(1):94–102. doi: http://dx.doi.org/10.1093/ during the ebola outbreak in liberia: the need for improvement in heapol/13.1.94 PMID: 10178189 emergencies. PLoS Curr. 2015 Apr 16;7:7. PMID: 25932347 26. Kruk ME, Rockers PC, Varpilah ST, Macauley R. Which doctor?: Determinants 12. Barden-O’Fallon J, Barry MA, Brodish P, Hazerjian J. Rapid assessment of of utilization of formal and informal health care in post conflict Liberia. ebola-related implications for reproductive, maternal, newborn and child Med Care. 2011 Jun;49(6):585–91. doi: http://dx.doi.org/10.1097/ health service delivery and utilization in Guinea. PLoS Curr. 2015 Aug 04;7:7. MLR.0b013e31820f0dd4 PMID: 21422954 PMID: 26331094 27. Mangham LJ, Hanson K. Scaling up in international health: what are the 13. Lori JR, Rominski SD, Perosky JE, Munro ML, Williams G, Bell SA, et al. A key issues? Health Policy Plan. 2010 Mar;25(2):85–96. doi: http://dx.doi. case series study on the effect of Ebola on facility-based deliveries in rural org/10.1093/heapol/czp066 PMID: 20071454 Liberia. BMC Pregnancy Childbirth. 2015 Oct 12;15(1):254. doi: http://dx.doi. 28. Smith JM, de Graft-Johnson J, Zyaee P, Ricca J, Fullerton J. Scaling up org/10.1186/s12884-015-0694-x PMID: 26459295 high-impact interventions: how is it done? Int J Gynaecol Obstet. 2015 14. Amouzou A, Morris S, Moulton LH, Mukanga D. Assessing the impact Jun;130 Suppl 2:S4–10. doi: http://dx.doi.org/10.1016/j.ijgo.2015.03.010 of integrated community case management (iCCM) programs on child PMID: 26115856 mortality: Review of early results and lessons learned in sub-Saharan 29. Yamey G. Scaling up global health interventions: a proposed framework for Africa. J Glob Health. 2014 Dec;4(2):020411. doi: http://dx.doi.org/10.7189/ success. PLoS Med. 2011 Jun;8(6):e1001049. doi: http://dx.doi.org/10.1371/ jogh.04.020411 PMID: 25520801 journal.pmed.1001049 PMID: 21738450 15. Mukanga D, Tiono AB, Anyorigiya T, Källander K, Konaté AT, Oduro AR, et al. 30. Rychetnik L, Frommer M, Hawe P, Shiell A. Criteria for evaluating evidence Integrated community case management of fever in children under five on public health interventions. J Epidemiol Community Health. 2002 using rapid diagnostic tests and respiratory rate counting: a multi-country Feb;56(2):119–27. doi: http://dx.doi.org/10.1136/jech.56.2.119 PMID: cluster randomized trial. Am J Trop Med Hyg. 2012 Nov;87(5 Suppl):21–9. 11812811 doi: http://dx.doi.org/10.4269/ajtmh.2012.11-0816 PMID: 23136274

120 Bull World Health Organ 2017;95:113–120| doi: http://dx.doi.org/10.2471/BLT.16.175513